Assessment Tools in Education

Just out: the new UNESCO report on AI and the future of education with a contribution from me and Dr Jasper Roe SFHEA. Our chapter: “𝑻𝒉𝒆 𝒆𝒏𝒅 𝒐𝒇 𝒂𝒔𝒔𝒆𝒔𝒔𝒎𝒆𝒏𝒕 𝒂𝒔 𝒘𝒆 𝒌𝒏𝒐𝒘 𝒊𝒕: 𝑮𝒆𝒏𝑨𝑰, 𝒊𝒏𝒆𝒒𝒖𝒂𝒍𝒊𝒕𝒚 𝒂𝒏𝒅 𝒕𝒉𝒆 𝒇𝒖𝒕𝒖𝒓𝒆 𝒐𝒇 𝒌𝒏𝒐𝒘𝒊𝒏𝒈” explores our prediction of the split between digitally advantaged and digitally marginalised contexts as GenAI tools continues to erode assessment practices. The first draft of our piece was described as “bleak” by one reviewer, and perhaps rightly so. We argue that traditional forms of assessment are collapsing under the pressure of GenAI and that the tools available today already challenge our ability to verify originality or authenticity. Detection of GenAI usage is unreliable, and even the exam hall is no safe harbour as we enter into a postplagiarism era (Sarah Elaine Eaton, PhD). Looking ahead, the risks are clear: • Digitally advantaged contexts may be able to experiment with new, AI-integrated assessments, moving towards human-centred skills like judgement, ethics, and collaboration. • Digitally marginalized contexts, however, risk being pushed back into outdated, rote-based exams, widening divides in opportunity and recognition. This raises urgent questions of power, equity, and epistemic justice: whose knowledge is validated, whose voices are sidelined, and how we build assessment systems that serve all learners. To prevent this split, we believe the following needs to happen: • Share AI infrastructure and training so all learners benefit. • Back multilingual and open-source AI to reduce language and cultural bias. • Invest in assessment practices that values relational expertise, ethical reasoning, and real-world impact. A big thank you to Shafika Isaacs (PhD) for her leadership on this, and Glen Hertelendy for the amazing organisation and production of the report. There are some huge names featured here, so it really is an honour to be among them. PS. If you are feeling depressed after reading our piece, read the chapter directly following ours 'The ends of tests: Possibilities for transformative assessment and learning with generative AI' by William Cope Mary Kalantzis and Akash Kumar Saini for an alternative perspective on the potential benefits that GenAI may bring to testing and assessment in the GenAI era.  

If AI can now produce competent answers in seconds, what exactly are we assessing in our degrees? AI is already embedded in how students learn, think, and produce work. So, the question is no longer about its use. Rather, the real question is whether assessment is designed to treat AI as a liability to be controlled or as a resource to be used well. AI-integrated assessment does not mean looking the other way when students use AI. It means designing tasks where AI use is expected, visible, and evaluated. The shift is subtle but fundamental: from policing outputs to assessing judgment. Several practical design principles follow. First, assess decisions rather than artefacts. In an AI-rich environment, polished outputs are cheap. What remains scarce is the ability to frame problems well, choose appropriate tools, test assumptions, and decide when not to trust an AI response. Assessment can require students to justify how AI was used, why particular prompts were chosen, and how outputs were validated against disciplinary knowledge. Second, make the process evidence assessable. Short AI logs, annotated iterations, or structured commentaries can document how thinking evolved through interaction with AI. This is forensic reasoning about choices made, alternatives rejected, and risks managed. Used well, it turns AI from a shortcut into a cognitive amplifier. Third, build in authentic constraints. In professional settings, AI is used within limits, including ethical rules, organisational policies, incomplete data, and reputational risk. Assessment can simulate these conditions through ambiguous briefs, imperfect datasets, or explicit governance boundaries. Students are evaluated on how they navigate trade-offs, not how elegant the final output appears. Fourth, reintroduce dialogue selectively. Ask for recorded walkthroughs or live critiques, which allow students to explain how AI shaped their reasoning. The purpose is not detection but sense-checking judgment. Weak understanding surfaces quickly when students must articulate why they trusted or rejected an AI-generated insight. Finally, reward responsible AI use explicitly. Rubrics should recognise transparency, validation, ethical awareness, and the integration of AI output with human judgement. When expectations are clear, students learn how to use AI well rather than how to hide it. This approach develops genuinely transferable skills such as judgment under uncertainty, learning agility, ethical reasoning, and accountability. It prepares students for workplace realities where AI is normal, governed, and consequential. It fosters better feedback and stronger academic relationships by shifting conversations from suspicion to reasoned discussion. The irony is that AI-integrated assessment is not easier. It is harder. It raises the bar. We need to shift our thinking from compliance to using assessment to develop graduates who not only know how to use AI, but also when, why, and to what effect.

The Michael J. Fox Foundation helped validate a biomarker for Parkinson's. Alpha-synuclein seeding amplification assay. Detects abnormal protein in spinal fluid. Before clinical symptoms appear. This is the Parkinson's biomarker we've needed. What it does: Finds misfolded alpha-synuclein - the "Parkinson's protein." In people diagnosed with Parkinson's: 88% positive. In people at high risk without symptoms: detects pathology years early. Like finding cancer Stage 1 instead of Stage 4. Why this matters: Parkinson's damage starts 10-20 years before tremor. By diagnosis, 60-80% of dopamine neurons already dead. Can't bring those back. But if we catch it earlier? When only 20% are damaged? Could slow or stop progression with the right drugs. The technology: Takes tiny amount of abnormal alpha-synuclein from spinal fluid. Amplifies it until detectable. Like PCR for COVID. Same principle. Binary result: pathology present or not. Who this helps now: Clinical trials can recruit earlier patients. Test disease-modifying drugs at stages where they might actually work. Current trials test drugs on people with 80% neuron loss. Of course they fail. Too late. With biomarker: test drugs on people with 20% loss. Actually have neurons left to save. What's coming: Optimizing the test to measure amount of pathology. Not just yes/no. How much. Track if treatments are working. See if protein levels decrease with therapy. Why now: New disease-modifying drugs in trials. Prasinezumab (anti-alpha-synuclein antibody) entering Phase 3. Ambroxol (boosts protein clearance) starting Phase 3 UK trial. Exenatide (diabetes drug) showing promise. All need early diagnosis to work. The diagnostic revolution: This is biology's century for Parkinson's. Move from clinical diagnosis to biological diagnosis. From "you have symptoms" to "you have pathology." Allows treatment before disability. The challenge: Spinal tap required. Not blood test yet. People hesitate. It's invasive. But researchers working on blood-based version. Alpha-synuclein appears in blood at lower levels. Harder to detect. But possible. Timeline: Biomarker validated now. Widespread clinical use: 3-5 years. Blood test version: 5-10 years. 💬 Would you want to know if you had Parkinson's pathology before symptoms? ♻️ Repost if early detection enables early treatment 👉 Follow me (Reza Hosseini Ghomi, MD, MSE) for biomarker breakthroughs that change medicine Citation: Siderowf A et al. α-synuclein seed amplification assay results from the Parkinson’s Progression Markers Initiative (PPMI) study. Lancet Neurol. 2023. Orrú CD, et al. Diagnostic and prognostic value of α-synuclein seed amplification assay kinetic measures in Parkinson’s disease including PPMI data. Lancet Neurol. 2025.

Moving away from thinking in AI as a "cheating" machine: The post discusses the updated version of the AI Assessment Scale (AIAS), a framework for integrating generative AI ethically into educational assessments across different disciplines. The AIAS provides five levels with varying degrees of permitted AI usage: 1. No AI: Students cannot use any AI tools. 2. AI-Assisted Idea Generation and Structuring: AI can be used for brainstorming and outlining, but final work must be human-authored.  3. AI-Assisted Editing: Students can use AI for refining and editing their work, submitting both original and AI-assisted content. 4. AI Task Completion, Human Evaluation: Students use AI for components of the task but critically evaluate the AI outputs. 5. Full AI: AI can be used throughout the task at the student/teacher's discretion. The updated AIAS aims to provide more nuance, flexibility and accommodate multimodal AI across diverse fields. Examples are given for applying each level to different assessment types. The author emphasizes the need to shift the narrative around AI in education from just "cheating" to exploring how it can enhance teaching and learning. The AIAS offers clarity to students on acceptable AI use and provides an ethical, equitable policy tool for institutions. The post includes an abstract from the published journal article further detailing the rationale and benefits of the AIAS framework. https://lnkd.in/ev-n_v4f

🔬 Which Lab Test for Which Disease? | Essential Guide for Clinicians & Students Accurate diagnosis begins with the **right laboratory investigation**. Understanding which test aligns with a specific disease can significantly improve patient outcomes and clinical decision-making. Here’s a quick, structured breakdown of common diseases and their recommended lab tests 🩸 Diabetes Mellitus ✔ Fasting Blood Sugar (FBS) ✔ Post-Prandial Blood Sugar (PPBS) ✔ HbA1c – best marker for long-term glycemic control (3 months) 🦋 Thyroid Disorders ✔ TSH – initial screening ✔ T3, T4 – to differentiate hypo/hyperthyroidism ✔ Anti-TPO Antibodies – for autoimmune thyroiditis (Hashimoto’s) ❤️ Heart Diseases ✔ ECG – arrhythmias & ischemia ✔ Lipid Profile – cardiovascular risk ✔ Troponin-I/T – acute myocardial infarction ✔ Echocardiography – valve function & pumping capacity 🧪 Kidney Diseases ✔ Kidney Function Tests (KFT) ✔ eGFR – staging of renal disease ✔ Urine routine + microscopy – proteins, casts ✔ Renal ultrasound – structural assessment 🟡 Liver Diseases ✔ Liver Function Test – bilirubin, SGOT, SGPT, ALP ✔ Ultrasound – fatty liver, cirrhosis ✔ Viral markers – HBsAg, Anti-HCV 🩸 Anemia ✔ Complete Blood Count (CBC) – MCV, MCH, MCHC ✔ Peripheral smear – morphology & type ✔ Iron profile / ferritin ✔ Reticulocyte count 🦠 Infections ✔ Typhoid – Widal / Typhidot / culture ✔ Tuberculosis – Mantoux, CBNAAT, CXR ✔ Malaria – peripheral smear / rapid diagnostic tests ✨ Bonus Tests • PCOD – Pelvic ultrasound, LH/FSH ratio, AMH • Rheumatology – RA factor, Anti-CCP, ANA • Cancer markers – PSA, AFP, CA-125 • General screening – CBC, RBS, LFT, KFT, lipid profile, Vitamin D/B12 📌 A reliable testing strategy leads to better diagnosis, timely intervention, and improved patient care.** Perfect for MLT students, clinicians, laboratory professionals, and healthcare educators. #MedicalLaboratoryTechnology #ClinicalDiagnostics #HealthcareProfessionals #Pathology #LaboratoryMedicine #MedicalStudents #Diagnostics #MLTCommunity #HealthcareEducation #ClinicalPractice

The American Psychiatric Association has acknowledged fundamental limitations in the current DSM. In its January 2026 roadmap, the APA states that the manual remains largely atheoretical, lacks validated biomarkers for most disorders, and depends on descriptive symptom clusters rather than established biological or etiological mechanisms. It proposes transitioning to a more flexible, digital format that incorporates contextual factors, transdiagnostic elements, severity gradients, and emerging scientific data. This development has direct implications for child welfare, family law, and trauma-informed practice, where DSM diagnoses have frequently supported child removals, termination of parental rights, and mandated interventions by framing trauma responses and systemic stressors as discrete disorders. A scientifically grounded alternative already exists and does not require waiting for a revised manual. Key components include: • Bruce Perry’s Neurosequential Model, which views distress as disrupted hierarchical brain development and sequences interventions from physiological regulation upward. • Jack Shonkoff’s ecobiodevelopmental framework, which demonstrates how toxic stress affects gene expression and developmental trajectories while protective relationships provide buffering effects. • The THEN Center’s emphasis on trauma—including structural inequities—as a primary driver of brain-body dysregulation. • Social and political determinants of health, which identify place-based inequalities, relational environments, and persistent unfreedoms as core influences on adaptive survival responses. These frameworks support multidimensional assessments focused on developmental history, neurobiological context, toxic stress exposure, protective factors, and systemic conditions. Existing tools such as Perry’s NMT Metric and ACEs/PCEs inventories provide practical alternatives that outperform DSM-based categorization. Child protection agencies should adopt differential response models, train personnel in trauma-informed and neurodevelopmental approaches, prioritize family support, and reduce reliance on DSM codes for decision-making. Attorneys should challenge DSM-dependent evidence in court under Daubert or Frye standards, request evaluations grounded in trauma and contextual factors, and advocate for reforms that uphold constitutional protections for families. The APA’s admission weakens the scientific justification for continued DSM dominance. The available evidence supports a shift toward systems that rely on valid science to address root causes rather than rely on outdated diagnostic constructs.

Navigating the Future of MRD & Liquid Biopsy in Oncology Trials   I wanted to give my thoughts on an insightful session at the 16th World Clinical Biomarkers & CDx Europe regarding the evolving role of Molecular Residual Disease (#MRD) testing and circulating tumor DNA (#ctDNA) by Jonathan Beer. John's hypothesis was that as we move toward precision medicine, selecting the right #LiquidBiopsy diagnostic strategy is critical for both trial execution and patient care.   Here are my key takeaways from his presentation:   John mentioned two ctDNA trials in his talk as exemplars for diagnostic strategy.   Firstly, resistance tracking in TRIDENT-1 where he described a diagnostic strategy prioritizing resistance mutation genes, that offers moderate sensitivity with speed, without fresh tissue.   Secondly, he discussed IMvigor010's retrospective analysis that revealed a treatment benefit for ctDNA+ subjects (HR 0.58). This lled to the landmark IMvigor011 prospective trial; which demonstrated how ctDNA testing could be used to personalise treatment for bladder cancer patients (MIBC) following surgery. It showed the superiority of serial testing and will likely lead to the first MRD #CDx approval.   He spoke about the need to tailor the assays to meet trial needs. If the timeline is short, a "fast" Tumor Naïve assay (TNA) is used, but when high-sensitivity is required, a Tumor Informed assay (TIA) is chosen instead. There are multiple parameters we need to consider, and whilst technical sensitivity is often the primary metric, many other factors count towards the right assay selection, from the availability of tissue/normal, the availability of other tests from the MRD vendor, e.g. CGP or HRD, or availability in a specific geography, i.e. China.   John was clear that we still need to work on breaking down adoption barriers. A significant hurdle for widespread clinical adoption of MRD testing outside the US has been our reliance on US-based central testing labs (Natera, Personalis, Inc., Myriad Genetics, et al).   I would suggest that the answer appears to be two-fold. Firstly, we need to develop decentralised solutions and secondly, we need those dominant labs to offer testing outside the USA, either by building labs or by partnering with centres of excellence (like Guardant Health's Marsden360 assay). John highlighted the collaborative development by Bristol Myers Squibb and Illumina on their tumor-informed WGS MRD workflow, which is being developed as a kit for any lab running NovaSeq. This is what I refer to as a Whole Genome Squared MRD (WGS2) as it uses WGS of FFPE tissue to identify variants for MRD detection in plasma WGS using a bioinformatic panel - it offers a logistically simpler solution that bespoke panel (dPCR, mPCR or HybCap) creation - the originators of this strategy were Dan-Avi Landau and Veracyte Inc. | C2i Genomics.   This shift towards faster, decentralized, and highly sensitive in-house MRD testing is a game-changer for oncology!  

I have never seen such drastic changes in university education as what has happened during the past two years because of generative AI technologies. Especially student assessment is now a completely different activity than what it used to be. I am starting to think that this requires a complete paradigm change in student assessments. We should not merely measure individual student capabilities but start evaluating student-AI teams and the result of the collaboration between AIs and students. Traditional university assessments are designed to measure individual student knowledge, skills, and critical thinking. Exams, essays, and projects typically emphasize personal effort and originality, aiming to cultivate independent thinkers. While this model has worked well for centuries, it now feels increasingly disconnected from the realities of the digital age. AI tools like ChatGPT, DALL-E, and others can produce sophisticated outputs, ranging from code and essays to data analysis and creative designs. Denying students access to these tools in assessments not only misrepresents their future work environments but also hinders their ability to develop critical skills for the AI-integrated workplace. The workplace of tomorrow will not reward individuals who can outperform AI but those who can work with AI to achieve exceptional outcomes. Universities must therefore adapt assessments to evaluate how well students integrate AI tools into their workflow to address complex, real-world problems, how critically they evaluate AI outputs for accuracy and bias, and how creatively and effectively they use AI to enhance their projects and generate novel solutions. Furthermore, students’ understanding of ethical considerations, including data privacy, transparency, and responsible innovation, must also become a focal point of assessment. Transitioning to a model that evaluates collaboration between students and AI requires innovative approaches. Assignments could explicitly require AI assistance, such as asking marketing students to develop campaigns with the help of AI tools, assess their viability, and justify their strategic decisions. Grading systems might prioritize the process over the final product, evaluating how students choose and use AI tools, iterate based on feedback, and address errors in AI-generated outputs. Open-book exams could allow AI use, with students evaluated on their ability to interpret, critique, and expand upon AI-generated content. Simulated workplace scenarios, where students work as part of a team with AI, could also become a powerful tool to measure real-world readiness. However, this transition is not without its challenges. See the comment section for more. Have you already started to assess the results of student-AI collaboration or do you still consider the individual capabilities of students as the main thing to assess in university education? #AI #education #assessment #grading #capabilities

How We Evaluate Technology at SOCPAC: A New Standard At SOCPAC, we’ve reached an inflection point in how we engage with technology companies. The days of buzzwords and slide decks are over. Moving forward, our evaluation process is guided by four criteria, each rooted in our operational needs and foundational architecture: 1. Production-Proven: Your technology must work in real-world environments, not just in a lab, demo, or wargame. If your product doesn’t already run at scale, on-network, and under pressure, it’s not ready for our missions. 2. User-Validated: We don’t just ask what your platform does. We ask: Do our operators want to use it? If an end user on our team says your tool gives them an edge, that carries more weight than any technical spec. 3. Architecture-Integrated: Every capability must connect to the platforms we’ve already deployed, a platform for strategic workflows and data fusion, a platform for tactical autonomy and sensor-to-shooter control, and a platform for AI tuning, feedback, and agent deployment. If your system can’t plug into this triad, it will create friction, not an advantage for us. 4. Culturally Aligned: We look for companies that embody intellectual honesty, speed of iteration, and a bias for solving problems over selling products. We want partners who thrive in ambiguity and innovate under constraint. This isn't about shutting the door. It's about raising the bar. We’re building a digital warfighting ecosystem, not a tech museum. If your team can plug into our architecture, align with our culture, and deliver capabilities that actually matter to the mission, we’re ready to work with you. Let’s move fast together.